Behind The Words

Twelve political speeches become a walk-through world of sound and light. Created with SCHNELLE BUNTE BILDER, the installation invites visitors to hear how the emotional tone of political language shifts, from joy to anger or fear.
‍

Listening to the emotion in political language.

Twelve perspectives.

The selection brings together six German politicians, three from elsewhere in Europe and three from the United States. Each transcript comes from an early speech in the politician’s latest campaign at the time the work was created. The rooms group the selection as left, centre and right.

Visitors standing between blue and red illuminated panels in Behind The Words — Behind The Words premiered at The Dark Rooms in Berlin, 2025.

The emotion in a speech becomes its musical score.

We turn AI estimates of a speech’s emotional tone into music. As that tone changes from paragraph to paragraph, sound and light shift together, giving each speech its own musical progression.

As visitors move among the illuminated panels, they encounter different emotional patterns through changing sound and light.

Listening becomes a way to explore political language side by side.

‍

Five visual treatments labelled fear, sadness, disgust, joy and anger — The five emotional categories share a musical and visual interpretation.

Five emotions. A shared musical vocabulary.

Joy, anger, sadness, disgust and fear each have a musical character shared across the twelve speeches. AI analysis estimates their presence and intensity, first across the whole transcript and then within each paragraph. This changing balance shapes the score; SCHNELLE BUNTE BILDER gives the same data visual form.

We developed the generative sound system in Max/MSP, using OSC to coordinate music and moving images. These are interpretations of language based on model estimates, rather than measurements of a speaker’s feelings.

Explore the speech analyses.

Choose a speech to follow its five emotional estimates in the upper graph. The lower graph adds estimates for hate speech, race-related stress and the plausibility of statements. Its fact-checking score is an approximate model assessment and needs independent verification.

← Previous Next →

Speech analysis for Emmanuel Macron: estimated emotional patterns across the speech.

Speech analysis for Gregor Gysi: estimated emotional patterns across the speech.

Speech analysis for Olaf Scholz: estimated emotional patterns across the speech.

Speech analysis for Alice Weidel: estimated emotional patterns across the speech.

Speech analysis for Donald Trump: estimated emotional patterns across the speech.

Speech analysis for Bernie Sanders: estimated emotional patterns across the speech.

Speech analysis for Jean-Luc Mélenchon: estimated emotional patterns across the speech.

Speech analysis for Marine Le Pen: estimated emotional patterns across the speech.

Speech analysis for Friedrich Merz: estimated emotional patterns across the speech.

Speech analysis for Christian Lindner: estimated emotional patterns across the speech.

Speech analysis for Kamala Harris: estimated emotional patterns across the speech.

Speech analysis for Robert Habeck: estimated emotional patterns across the speech.

No items found.

Four speeches share each room. As visitors move among the illuminated panels, they encounter different emotional patterns through changing sound and light.

Walk between political voices.

Listening becomes a way to explore political language side by side.

Visitors silhouetted between illuminated panels in the exhibition space — Visitors among the sound and light panels at The Dark Rooms, Berlin.

Project Scope

Inviting visitors to listen beyond the words of political speeches, through music shaped by their emotional patterns.

Explore project scope ↗

Function of Sound

Data sonification.

We make the emotional content of political speeches audible. By translating the analysis into music, we let audiences encounter the tone and emotional shifts of the language through listening.

Sound Experience

Walk between emotional portraits.

Each speech occupies a position in the room. Visitors encounter its shifting emotional character through music and synchronised light as they move among the panels.

Content

5

emotions, a shared musical palette.

Joy, anger, sadness, disgust and fear each have a musical character shared across all twelve speeches. Their estimated balance changes paragraph by paragraph, shaping a distinct musical progression for each transcript.

Sound & Space

Political voices, side by side.

Twelve LED panels with integrated speakers are arranged across three rooms, four per room. Each speech has its own position within the shared sound field, placing the different political voices side by side.

Studio Contribution

A generative sound system.

Our contribution combined sound concept, data sonification, composition and programming of the generative sound system in Max/MSP. OSC connected the analysis with the audiovisual score. SCHNELLE BUNTE BILDER created the visuals for the jointly developed artwork.

Credits

ArtworkKLING KLANG KLONG in collaboration with SCHNELLE BUNTE BILDER
VisualsSCHNELLE BUNTE BILDER
Premiere / yearThe Dark Rooms · Berlin · 2025
CameraFrank Sauer
Sound experience designKLING KLANG KLONG